Microbiota variations in Culex nigripalpus disease vector mosquito of West Nile virus and Saint Louis Encephalitis from different geographic origins

Although mosquito microbiota are known to influence reproduction, nutrition, disease transmission, and pesticide resistance, the relationship between host-associated microbial community composition and geographical location is poorly understood. To begin addressing this knowledge gap, we characterized microbiota associated with adult females of Culex nigripalpus mosquito vectors of Saint Louis Encephalitis and West Nile viruses sampled from three locations in Florida (Vero Beach, Palmetto Inland, and Palmetto Coast). High-throughput sequencing of PCR-amplified 16S rRNA genes demonstrated significant differences among microbial communities of mosquitoes sampled from the three locations. Mosquitoes from Vero Beach (east coast Florida) were dominated by uncultivated Asaia sp. (Alphaproteobacteria), whereas microbiota associated with mosquitoes collected from two mosquito populations at Palmetto (west coast Florida) sites were dominated by uncultured Spironema culicis (Spirochaetes), Salinisphaera hydrothermalis (Gammaproteobacteria), Spiroplasma (Mollicutes), uncultured Enterobacteriaceae, Candidatus Megaira (Alphaproteobacteria; Rickettsiae), and Zymobacter (Gammaproteobacteria). The variation in taxonomic profiles of Cx. nigripalpus gut microbial communities, especially with respect to dominating taxa, is a potentially critical factor in understanding disease transmission and mosquito susceptibility to insecticides among different mosquito populations.

wetlands (O'Meara & Evans, 1983;Darsie & Ward, 2005). This species typically represents the dominant Culex colonizer of newly formed aquatic habitats (Duguma et al., 2017a). Genetic variations within populations of Cx. nigripalpus vary in time and space (Nayar, Knight & Munstermann, 2002), including tolerance to organophosphate-based adulticides that occurs in some populations (Boike et al., 1989;Shin & Smartt, 2016). Although variations of genetic factors may largely be responsible for variations in susceptibility to pathogens and pesticide tolerances, the role of non-genetic factors, such as microbial symbionts found in different Cx. nigripalpus populations, is not well understood.
The microbiota of mosquitoes can help provide nutrition for successful development (Mitraka et al., 2013;Coon et al., 2014), influence pathogen transmission (Glaser & Meola, 2010;Finney, Kamhawi & Wasmuth, 2015;Zink et al., 2015;Shaw et al., 2016;Novakova et al., 2017), and impact susceptibility to pesticides (Patil et al., 2013;De Almeida et al., 2017;Dada et al., 2018). Some mosquito-associated microorganisms have also been implicated in degrading insecticides, such as organophosphates, and may enhance mosquito tolerance to pesticides (De Almeida et al., 2017;Soltani et al., 2017;Dada et al., 2018). Dada et al. (2018) reported a dominance of bacteria associated with organophosphate-degrading enzymes in organophosphate resistant malaria vector mosquito, Anopheles albimanus Wiedemann suggesting a link between insecticide resistance and microbiota composition. Mosquitoes devoid of microbiota (using antibiotics) were more susceptible to Bacillus thuringiensis compared to mosquitoes that contain natural microflora (Patil et al., 2013). The success of a Wolbachia-based biocontrol program, which is dependent on the success of colonization of Wolbachia strains into mosquito populations that lack those strains, has also been reported to be potentially limited by larval environmental factors (Hancock et al., 2016). Possible effects of native microbial communities present in various larval habitats, and ingested by larval mosquitoes, naturally occurring in different mosquito populations may influence the success of this type of control program. Wolbachia and other microbiota in mosquitoes varied across seasons in WNV mosquito vectors and Wolbachia infections were found negatively correlated with WNV in these mosquitoes (Novakova et al., 2017). Microbiota in naturally occurring mosquitoes are known to differ from laboratory-reared mosquitoes (Duguma et al., 2015), and are likely to influence mosquito vector competence both seasonally and locally (Novakova et al., 2017). Spatial and temporal environmental variations influence the microbiota of insects, including fruit flies and mosquitoes (Corby-Harris et al., 2007), and can alter susceptibility to pathogen infections (Akorli et al., 2016;Novakova et al., 2017).
To better investigate the roles of microbiota in Cx. nigripalpus, first it is important to understand biogeographical patterns for microbial communities within unique Cx. nigripalpus field populations. By sampling adult female mosquitoes from three different geographical locations in Florida, the goal of this research was to identify core microbial taxa, despite geographical distance, in addition to microorganisms that are uniquely distributed spatially. In a previous study, mosquito microbiota composition in general did not vary despite differences in environmental variables such as the amount of nutrients available in the developmental habitat (Duguma et al., 2017a). Microbiota from mosquitoes originating from mesocosms placed in the same geographic location with high and low nutrient concentration in water column did not differ significantly (Duguma et al., 2015(Duguma et al., , 2017a. As a result, here, we hypothesized that adult Cx. nigripalpus mosquito populations sampled from different geographic locations would not differ in their gut microbiota. Characterizing the microbiota associated with Cx. nigripalpus will help identify potential symbionts and known pathogens that can be used to develop novel approaches for mosquito control, predict variations in susceptibility to different pesticides, and monitor pathogen infections among different populations.

Sampling
Adult females of Cx. nigripalpus were sampled from three coastal habitats of Florida in autumn 2015, including the Institute of Food and Agricultural Sciences Florida Medical Entomology Laboratory (FMEL; University of Florida, Gainesville, FL, USA) outdoor aquatic mosquito experiment station in Vero Beach (east coast Florida; 27.5876 N, -80.3735 W) and two sites from Palmetto located on the west coast of Florida (Fig. 1). The two Palmetto sites are referred to as "Palmetto Coast" (27.6405 N,. Although both Palmetto locations are over 200 km west of the Vero Beach site, Palmetto Coast is located about one km from coastal habitats of the Gulf of Mexico and Palmetto Inland is located about six km southwest of the Palmetto Coast site and is closer to residential areas. Samples were collected in October 2015 and December 2015 from Vero Beach and Palmetto sites, respectively. Adult mosquitoes at the Vero Beach site were sampled following a previously described procedure (Duguma et al., 2017a). Late instar larvae were taken from four outdoor aquatic mesocosms located at the FMEL campus and placed into four modified BioQuip mosquito-rearing jars (BioQuip, Inc., Rancho Dominguez, CA, USA) for adult emergence where adults were collected as soon as they emerged. The mesocosms were constructed using cattle tanks (volume = 378 L) and placed in the field environment surrounded by oak trees to simulate natural breeding habitats for Culex species including Cx. nigripalpus. The trees surrounding mesocosms provided continuous supply of organic substrates, source microbes, and mosquitoes. The modified BioQuip mosquito-rearing jars float in the mesocosms with the larvae placed in the mesh screen in the water column allowing mosquito larvae access to microbial food sources (Duguma et al., 2017a). Culex nigripalpus larvae were identified using the keys before placed in BioQuip mosquito-rearing jars (Cutwa & O'Meara, 2006). Upon adult emergence, female adults were euthanized and placed in freezer until DNA extraction. Together, four samples (one sample per mesocosm) were collected from Vero Beach site. Vero Beach mosquitoes did not feed on blood or sugar post emergence. The mosquitoes from Palmetto sites were collected using CDC Miniature light traps (John W. Hock Company, Gainesville, FL, USA) with dry ice as bait. These mosquitoes were collected from Manatee County Mosquito Control District operational sites, ensuring exposure of at least parents of the analyzed mosquitoes to insecticides. However, no attempt was made to detect presence of pesticide resistance in mosquitoes. A total of four trap samples, two from each of Palmetto Inland and Palmetto Coast were collected. Mosquitoes were identified to species and sorted by their sexes microscopically, kept in 95% ethanol, and stored in a freezer until DNA extraction and further processing.
Together, 16 mosquito samples, including four from the Vero Beach site and 12 from the two Palmetto sites, representing east and west coast mosquito populations were analyzed for their microbial community profiles. A sample consisted of a pool of three intact individual female adult mosquitoes used for DNA extraction. The mosquitoes from Vero Beach were captured and euthanized following eclosion. However, we have no information on the age of mosquitoes from Palmetto sites because they were captured in CDC light traps. Although mosquitoes collected and used from Palmetto sites were not blood-fed ones, we have not excluded the possibility that they may have fed on vertebrates in their life time.

DNA extraction, PCR, and sequencing
Surface sterilization by 95% ethanol followed by triple rinsing using molecular biology grade UltraPure water (Quality Biological, Inc., Gaithersburg, MD, USA), followed by air drying was performed before DNA extraction. The DNA extraction, PCR, and sequencing procedures were described in previous studies (Bartram et al., 2011;  Duguma et al., 2015Duguma et al., , 2017a. The DNA from the whole body of mosquitoes was extracted using DNeasy Blood and Tissue Kit following the manufacturer's protocol (Qiagen, Inc., Valencia, CA, USA) under a laminar flow hood. We only extracted DNA from adult females to focus on them and identify bacterial symbionts associated with females and likely to be maternally inherited. Amplicons of approximately 460 base pairs were generated by PCR of V3 and V4 regions of 16S rRNA genes using Pro341F (5′-CCTACGGGNBG CASCAG-3′) and Pro805R (5′-GACTACNVGGGTATCTAATCC-3′), targeting both bacteria and archaea (Takahashi et al., 2014). Amplicons from each of the samples were tagged with six-base barcodes, amplified using Illumina-specific primers, pooled, and subjected to 250-base paired-end sequencing (Reagent Kit V2, 500) on a MiSeq (Illumina Inc., San Diego, CA, USA).

Data analyses
Sequence analysis was performed using QIIME2 (2018.6 release) with additional analyses performed using R in Jupyter notebooks. Divisive amplicon denoising algorithm 2 (version 1.6.0) was used to generate the amplicon sequence variants (ASVs) from the raw sequences and detect and remove chimeric sequences (Callahan et al., 2016). The consensus sequences for the ASVs were classified with a scikit-learn Naïve Bayes classifier trained against the most recent SILVA 16S rRNA gene reference (release 132) database (Quast et al., 2012). Sequences were rarefied to the lowest number of sequences per sample (31,183) for alpha diversity analyses. Non-metric multidimensional scaling based on the weighted and unweighted (i.e., binary and non-binary) Bray-Curtis dissimilarities was performed using R package "vegan" version 2.5-2 (Oksanen et al., 2007). From the same package, the multi-response permutation procedure (MRPP) was used to test differences among sample groups. Faith's phylogenetic diversity (Faith, 1992) was computed with QIIME2 by using the generated phylogeny as input (Katoh & Standley, 2013) and phylogenies were created by FastTree version 2.1.10 (Price, Dehal & Arkin, 2010). Faith's phylogenetic diversity (Faith, 1992) alpha-diversity measure was computed with QIIME2, using the generated phylogeny as input. Differences in alpha diversity measured by phylogenetic diversity were determined using Kruskal-Wallis followed by Dunn's post hoc test with Benjamini-Hochberg multiple hypothesis test correction. Upset plots were used to determine overlaps of ASVs and sequences from mosquitoes sampled from the three locations (Conway, Lex & Gehlenborg, 2017). The relative abundance plot was generated using the full data set, whereas the ordinations plots were based on rarefied data. Samples from this study were sequenced along with samples analyzed by previous publications (Duguma et al., 2017a(Duguma et al., , 2017b, and 24 non-template negative controls were used in the same sequence run. Downstream analysis of the sequences revealed that our biological field samples were distinct from the very few microbial taxa associated with negative controls, suggesting that the mosquito sample data we analyzed were not affected by laboratory or reagent contamination. In addition, the non-template samples had fewer sequences (maximum of 16,000 sequences in negative control samples, versus a minimum of 34,286 sequences in field samples). The negative controls were dominated by Facklamia, Staphylococcus epidermidis, Vibrio metschnikovii, and Actinomycetales, and were removed from further analysis and interpretation.
The sequence dataset generated for this study was submitted to the European Bioinformatics Institute under project accession number PRJEB24029. Data and notebooks are available on GitHub (https://github.com/mwhall/PeerJ_Culex_nigripalpus). Scripts for this publication have been archived in release 1.0 (DOI 10.5281/zenodo.1484552).

RESULTS
To explore the microbiota of wild adult female Cx. nigripalpus mosquitoes sampled from three field locations in Florida (Fig. 1), we generated a total sequence dataset of 1,583,222 16S rRNA gene sequences that clustered into 1,482 ASVs, including 1,198 bacteria, 257 unclassified ASVs, 25 Eukaryotes, and two archaea. Together, the bacterial ASVs were affiliated with 22 unique phyla. Proteobacteria was the dominant phylum associated with adult female Cx. nigripalpus, with site-specific average relative abundances ranging between 35.3% and 99.7% among samples from the three locations (Table 1). In samples that contained lower relative proportions of Proteobacteria, ASVs associated with Spirochaetes and Tenericutes dominated Cx. nigripalpus microbiota at >10% relative abundance.
Ordination analysis using a weighted Bray-Curtis distance measure revealed consistent differences in microbial communities among mosquitoes collected from the three sample site locations ( Fig. 2A; MRPP test; A = 0.19; p = 0.001). Similarly, an ordination based on unweighted Bray-Curtis (i.e., Sørensen-Dice) distances revealed that microbiota from wild Cx. nigripalpus mosquitoes collected from Vero Beach and two Palmetto sites (Palmetto Inland and Palmetto Coast) differed significantly, but with a smaller effect  The mosquitoes from the Palmetto Coast site contained significantly higher microbial diversity than the Vero Beach site ( Fig. 2C; Dunn's test; p = 0.02). The Palmetto Inland mosquitoes had a higher median phylogenetic diversity than the Vero Beach mosquitoes, but this difference was not statistically significant ( Fig. 2C; Dunn's test; p = 0.104). At the ASV level of resolution, there were no ASVs found in all samples. The majority of ASVs were observed only in one location (Fig. 2D). However, there were 10 ASVs observed in all three sites (Fig. 2D) which accounted for 654,848 sequences and represented 41.4% of all sequences (Fig. 2E). This included an abundant ASV of Spironema culicis, two abundant ASVs of Asaia, and seven relatively low abundance (<0.1%) but nonetheless frequently present ASVs, including Thorsellia, Arcobacter, Rubritepida, and uncultured Erysipelotrichaceae.
One ASV classified to Asaia sp. of the Acetobacteraceae was the dominant taxon from the Vero Beach site, accounting for 56.1-81.0% of all sequences (Fig. 3). This ASV,    (Fig. 3).

DISCUSSION
Many factors including sex, age, size, geographic location, and other physiological status of mosquitoes have been shown to influence mosquito-associated microbiota (Osei-Poku et al., 2012;Zouache et al., 2010;Buck et al., 2016;Minard et al., 2018). Our study corroborated these previous findings and indicated that factors associated with different geographic locations might have impacted microbiota composition variations observed in adult female Cx. nigripalpus collected from different geographic locations. It is possible that the larval environment of the adult mosquitoes collected from the different sites and the physiological statuses of the mosquitoes including age and parity may have contributed to this variation (Akorli et al., 2016). The mosquitoes from Vero Beach were collected from a relatively similar larval site; however, we are uncertain of the specific larval habitats of samples collected from the two Palmetto sites. The samples from Vero Beach were collected from mesocosms located in one location. In contrast, because mosquitoes from Palmetto sites were trapped as adults in CDC light traps, we could not exactly describe their larval developmental origin. While our study has not accounted for the possible effect of age, size, and other physiological statuses, distinct microbiota in the mosquitoes might reflect differences in environmental factors associated with larval developmental sites or geographic locations. Using baseline data from this study, future research will help disentangle whether variations in microbiota among samples observed in this study may be due to factors such as local microclimatic differences in mosquito development or physiological status. Proteobacteria dominated microbiota in Cx. nigripalpus mosquitoes (this study; Duguma et al., 2017a) similar to studies on different mosquito species (Duguma et al., 2013;Minard, Mavingui & Moro, 2013;Dada et al., 2014). It is currently unknown whether the dominance of Proteobacteria members may have some inhibitory effects whether through competition inside the host to other bacterial taxa, but our studies show that Cx. nigripalpus samples that contained higher proportion of Proteobacteria contained lower proportion of Spirochaetes, Tenericutes, and Firmicutes.
The ASVs classifying to the alphaproteobacterial genus Asaia dominated all of the Vero Beach samples (Fig. 3). The sequences identified to Asaia in this study were the same sequences identified to Swaminathania in previous studies (Duguma et al., 2017a(Duguma et al., , 2017b, but were not observed in such high relative abundances in the previous studies. Asaia spp. symbionts have previously been linked to several medically important mosquitoes, including Anopheles malaria vectors, with some suggesting a critical role for the Asaia bacteria in the complete development of Anopheles mosquito species (Chouaia et al., 2012;Mitraka et al., 2013) and Aedes mosquitoes (Favia et al., 2007;Damiani et al., 2010). These bacteria were also detected in four species of the Cx. pipiens Linaeus complex, including Cx. quinquefasciatus Say (De Freece et al., 2014). Asaia species have been reported to outcompete and impede the proliferation of Wolbachia in both laboratory-reared and field-collected mosquitoes (Hughes et al., 2014;Rossi et al., 2015;Novakova et al., 2017). Because of the Asaia association with both laboratory and field-collected mosquitoes, and reported anti-Plasmodium properties, this bacterial species has been considered for paratransgenic control of mosquitoes (Wilke & Marrelli, 2015). Our finding of variations of Asaia occurrence in Cx. nigripalpus vector populations warrants further investigation, including sequencing the genome of this bacterial strain. Further investigation is warranted to determine the proliferation of Asaia taxa in some mosquito samples and whether its dominance is correlated with the scarcity of Wolbachia in Cx. nigripalpus (Duguma et al., 2017a(Duguma et al., , 2017b, as has been suggested for Anopheles mosquitoes (Rossi et al., 2015).
A gammaproteobacterial genus Salinisphaera was detected at >10% relative abundance in four Palmetto Inland sites, and species of this genus are associated with habitats in the transition zone between freshwater and saltwater, and known to produce detoxification enzymes involved in the degradation of pesticides (Vetriani, Crespo-Medina & Antunes, 2014). It was found in a recent study to be the only genus to increase in relative abundance during a Wolbachia infection of Aedes mosquitoes (Audsley et al., 2018). Bacterial communities associated with insecticide-resistant mosquitoes differed from those of susceptible mosquitoes, with a greater relative abundance of Firmicutes and Actinobacteria found in organophosphate-resistant mosquitoes as compared to susceptible mosquitoes (Dada et al., 2018). Burkholderiales members are associated with degradation of organophosphate insecticides (fenitrothion) in other insects (Kikuchi et al., 2012), whereas, their role in mosquitoes has yet to be determined. Although we have not determined whether mosquitoes sampled and analyzed from the Palmetto areas were exposed to pesticides in this study, a previous study showed that mosquitoes from west coast Florida including the Palmetto area might have been associated with pesticide resistance (Shin & Smartt, 2016).
The majority of samples from Palmetto sites contained an ASV of Spiroplasma (Mollicutes), but specifically dominated bacterial communities in two samples ( Fig. 3; PI2 and PC4). Spiroplasma species have been previously isolated from several mosquitoes, including Cx. nigripalpus (Shaikh et al., 1987), and are considered mosquito pathogens (Humphery-Smith, Grulet & Chastel, 1991a;Humphery-Smith et al., 1991b;Chang et al., 2014). Some strains of these bacteria have also been linked to killing males of fruit fly Drosophila melanogaster while providing immunity against parasitism by other insects (Xie et al., 2014). It is currently unknown what role it might play in adult Cx. nigirpalpus mosquitoes. An ASV classifying to candidatus "Spironema culicis", another spiral-shaped bacterium, also dominated bacterial communities in three samples from Palmetto sites (PI3, PC5, and PC6). This bacterium has been previously found in Cx. pipiens (Čechová et al., 2004), but has not been previously reported to be able to dominate the mosquito gut.
The consistent detection of Thorsellia spp., although low in abundance, corroborates previous findings of their close association with cross genera mosquito vectors of WNV and SLE viruses (Duguma et al., 2015(Duguma et al., , 2017b and malaria parasites (Lindh, Terenius & Faye, 2005;Rani et al., 2009). Although mosquito midgut bacteria usually proliferate post-blood feeding or after sugar meals (Pumpuni et al., 1996), detection of Thorsellia spp. in newly emerged mosquitoes, as well as host-seeking adults, suggests a strong association of these bacteria with Cx. nigripalpus mosquitoes. These bacteria have been more prevalent in larval mosquitoes (Duguma et al., 2017b) suggesting that they were likely environmentally acquired.
Mosquitoes analyzed in this study were either newly emerged and not blood-fed (Vero Beach mosquitoes) or were seeking hosts at the time they were collected (Palmetto mosquitoes) and were not expected to acquire pathogens from vertebrate blood. However, low proportions of some bacterial ASVs associated with animal and human pathogens or parasites, including B. cereus, Rickettsiella sp. Vibrio cholerae, and Xanthomonadaceae, were found in some Cx. nigripalpus samples collected from Palmetto sites suggesting that mosquitoes from these sites may have been in contact with vertebrate hosts prior to their collection.
Despite originating from the same mosquito species, life stage, or same geographic region, we observed a remarkably high heterogeneity in the taxonomic profiles of Cx. nigripalpus microbiome (Fig. 3). On one hand, there was a tendency of some mosquitoes associating with uneven bacterial species distributions containing one or two ASVs with very high abundance that account for 90% or more of all sequences. The species that showed the tendency to dominate bacterial communities in Cx. nigripalpus include known symbionts such as Asaia (this study) and Wolbachia (Fig. 3, Sample 12; Duguma et al., 2017a). However, this has to be taken with caution and further investigation using Asaia and Wolbachia gene-specific primers is needed to validate this observation. Other dominant group include Spironema culicis, Salinisphaera hydrothermalis, an unclassified Enterobacteriaceae ASV, Candidatus Megaira, an uncultured Burkholderiaceae, and to a lesser degree, Pseudomonas and Aeromonas (Fig. 3). The high representation of these taxa in certain mosquito samples but not in others warrant further investigation to understand the underlying biological factors causing this variation.
On the other hand, a relatively high bacterial ASV evenness was observed in many of the samples (Fig. 3). It is evident that many taxa are capable of dominating the mosquito gut environment. The causes or consequences of a high or a low diversity of gut microbial community are not clear. However, in order to harness the gut microbiota for disease vector control or understand their potential impact on pesticide resistance, the heterogeneity in microbial communities in the same species underscores the importance of elucidating the ecological, environmental, genetic, and evolutionary factors that are involved in the assemblage of the mosquito gut community. Our study provides a baseline information to further elucidate the impact of resident microbial taxa in naturally occurring mosquitoes collected from different localities to design symbiotic control such as Wolbachia-based vector and disease control approaches.

CONCLUSIONS
Our study indicated microbiota variations among mosquito populations in Cx. nigripalpus mosquito disease vector of SLE and WNV. Future research including mosquitoes from broader geographic regions and seasons will assist in better understanding the consequences of microbiota variations in mosquito disease transmission ability and pesticide tolerance (if any) in Cx. nigripalpus. Further investigation in examining the factors for the tendency for some taxa to dominate the mosquito gut is required, as they may outcompete taxa such as Wolbachia when used as vector-borne disease vector control.